Skip to content

v1.0.1 - AI Model Tiering & Automated Adaptation

Choose a tag to compare

View all tags
@andyleimc-source andyleimc-source released this 20 Mar 01:40
· 354 commits to main since this release
v1.0.1
d18bdbf

🚀 Major Refactor: Intelligent Model Tiering

This release introduces a task-driven AI model selection logic, significantly improving generation quality while maintaining execution efficiency.

Key Changes:

  • Task Tiering (REASONING vs FAST):
    • REASONING Tier: Used for complex architectural planning (Worksheets, Workflows, Layouts, Roles).
      • Maps to deepseek-reasoner (R1) or gemini-2.5-pro.
    • FAST Tier: Used for high-volume or straightforward tasks (Mock Data, Icons, Basic Views).
      • Maps to deepseek-chat (V3) or gemini-2.5-flash.
  • Zero-Config Selection: Users no longer need to pass --model arguments. The system automatically selects the best model based on the provider configured in ai_auth.json.
  • Exhaustive Cleanup: Removed legacy DEFAULT_MODEL, GEN_MODEL, and GEN_API_KEY hardcoding across 30+ scripts.
  • Improved Robustness: Centralized configuration loading in ai_utils.py with better error handling and fallback logic.

Verified Pipelines:

  • DeepSeek Line: Full end-to-end success using Reasoner + Chat combo.
  • Gemini Line: Full end-to-end success using 2.5 Pro + Flash combo.
Assets 2
Loading